AppsFlyer Improved BI Performance and Simplified Analytics with Starburst

About

AppsFlyer is a global mobile measurement and analytics platform that helps businesses understand how users discover, engage with, and behave across mobile applications.

Internally, AppsFlyer maintains years of historical business data and relies heavily on analytics to support sales, marketing, customer success, and strategic decision-making. Looker serves as the primary analytics interface for approximately 50–60 analysts across the organization.

Challenge

AppsFlyer’s Data and AI Engineering team is responsible for managing the company’s business data platform, integrating systems such as Salesforce, Marketo, Mixpanel, Jira, Datadog, and other business tools.

Before Starburst, Amazon Athena powered Looker’s BI dashboards. As data volumes grew and usage increased, performance challenges began to impact day-to-day operations:

• BI dashboards became slow, particularly with complex queries and large datasets
• Analysts relied on Looker daily to monitor tickets, performance metrics, and customer trends, making performance critical
• Data had to be copied and duplicated across platforms (including Athena and BigQuery) to perform joins
• ETL complexity increased engineering workload and introduced potential data discrepancies

Slow dashboards directly affected analysts, who in turn supported sales, marketing, and customer-facing teams. Performance issues required ongoing attention from the data engineering team and created friction in daily operations.

Solution

AppsFlyer introduced Starburst Galaxy to replace Athena as the primary query engine for Looker and to serve as a federated query layer across multiple data platforms.

Starburst delivered value in two key areas:

1. Improved BI Performance
Starburst significantly improved query performance for Looker’s BI dashboards. After deployment, complaints about slowness largely disappeared, and analysts were able to work more efficiently with large datasets.

2. Federated Querying Across Platforms
AppsFlyer stores data across multiple platforms. Starburst enables cross-catalog joins and federated queries across these environments without copying or duplicating datasets.

This simplified the data architecture, reduced ongoing engineering effort, and minimized discrepancies between systems — allowing the team to focus on delivering insights rather than managing pipelines.

Results

Following the adoption of Starburst Galaxy, AppsFlyer experienced measurable operational improvements:

Faster, More Reliable Dashboards
BI performance improved significantly. The team no longer receives regular complaints about slowness, and analysts can work more efficiently with large datasets.

Increased Analyst Velocity
With faster queries and simplified access to data across platforms, analysts can build more models and generate insights more efficiently.

Reduced Data Engineering Overhead
By eliminating data duplication and reducing cross-platform ETL complexity, the data engineering team reduced maintenance burden and simplified architecture.

Improved Operational Efficiency
AppsFlyer achieved cost efficiencies by reducing engineering time, simplifying data management, and reducing future risks from discrepancies and pipeline failures.